Machine learning for turbulence modeling

In this tutorial, we offer an overview of different machine learning (ML) methodologies for turbulence modeling. The main focus is on RANS and LES but other reduced-order models are also discussed. First, the distinction between models and methods is highlighted. The different ML approaches are introduced according to three classifications: i) the level of modeling form of closure terms, ii) machine learning methodology and, iii) the neural network architecture. A selected number of works are used to illustrate the ML approaches. Finally, we address the question: What can we do with machine-learning models for turbulence that we couldn’t do before?

Presentation: Video